中国科学院沈阳自动化研究所机构知识库
Advanced  
SIA OpenIR  > 光电信息技术研究室  > 学位论文
题名: 基于GPU的实时图像处理方法研究
其他题名: Research on Real-time Image Processing Based on GPU
作者: 刘正斌
导师: 朱丹
分类号: TP391.41
关键词: GPU ; 结构光 ; 缺陷检测 ; 运动估计
页码: 55页
学位专业: 模式识别与智能系统
学位类别: 硕士
答辩日期: 2013-05-28
授予单位: 中国科学院沈阳自动化研究所
作者部门: 光电信息技术研究室
中文摘要: GPU(Graphic Processing Unit)是一种市场上很容易获得的低成本、高性能处理器,拥有高度并行性和可编程性的GPU已经广泛应用于各种通用计算领域。本论文利用GPU来实现结构光三维视觉测量和电子稳像系统中的运动矢量估计。结构光三维视觉测量技术因其非接触、动态响应快、精度高等优点被广泛应用于各种表面检测领域。采用结构光视觉对其进行缺陷检测需要占用大量的计算资源,往往难以满足实时检测的要求。电子稳像技术是通过数字图像处理技术对序列图像进行运动估计,进而进行运动补偿来消除图像帧间的诸如抖动、旋转等非正常偏移的一种技术,其中的关键技术是全局运动矢量估计。在整个电子稳像系统中,运动矢量估计需要非常大的计算量,消耗大量的计算资源,几乎占用了整个电子稳像系统大部分的计算时间,因此提高运动矢量估计的计算效率成了提高电子稳像系统实时性的关键。本文首先对基于结构光视觉的钢轨表面缺陷检测作了深入研究,提出了一种结构光中心的快速提取算法——Steger算法的改进算法,并将该算法在GPU中实现。对于钢轨表面缺陷深度和宽度的计算,本文则提出了一种基于形态学的计算方法,并在静态情况下对钢轨表面缺陷检测进行实验,实验结果显示,GPU对于结构光中心的快速提取具有非常好的效果。同时,本文对于运动矢量估计算法进行研究,其中以块匹配算法及其改进算法作为重点研究对象,对于块匹配算法的一种改进算法——GEA算法(Global Elimination Algorithm),进行并行化处理,并利用GPU实现该算法。针对具有平移和小的旋转抖动的航拍视频进行实验,实验结果表明,GPU对于运动矢量估计具有非常好的加速效果,能够满足运动矢量估计的实时处理。
英文摘要: GPU (Graphic Processing Unit) is a kind of low cost and high performance processor that is easy to get from market, and GPU with high parallelism and programmability has been widely used in all kinds of general-purpose computing field. In this paper, GPU is used to implement structured light 3d vision measurement and motion vector estimation. Structured light 3d vision measuring technique has been widely used in various surface defects detection field because of its advantages of non-contact, fast dynamic response and high precision. Defects inspection using structure light vision takes up a lot of computing resources, and it is often difficult to meet the requirement of real-time detection. Electronic image stabilization is a technology that uses the digital image processing technology to estimate the motion vectors on image sequences and then compensate them to remove the non-normal movement such as image jitter and rotation, and the key technology is the global motion vector estimation. Throughout the electronic image stabilization system, motion vector estimation requires a very large amount of calculation, consumes large amounts of computing resources, and almost occupies most of the computational time of the whole electronic image stabilization system. Thus improving the computing efficiency of motion vector estimation is the key to real-time electronic image stabilization system. Firstly, in this paper,an intensive study is made on rail surface defects detection based on structure light vision, a fast algorithm of structure light center extraction -- the improved algorithm of Steger algorithm, is put forward and its parallel algorithm is realized in GPU. For the calculation of the depth and width of the rail surface defects, a calculation method based on morphology is put forward in this paper. Under the static condition, the experiment of rail surface defects detection is made, and the result shows that GPU has a very good effect in extracting structured light center. At the same time, this paper studies motion vector estimation algorithm, with block matching algorithm and its improved algorithm the key research object. As an improved algorithm, GEA algorithm (Global Elimination Algorithm) is paralleled processed and is implemented using GPU. Aerial video with translation and small jitter is used in the experiment, and the result shows that GPU has very good speedup for motion vector estimation, and can meet the real-time processing of motion vector estimation.
语种: 中文
产权排序: 1
内容类型: 学位论文
URI标识: http://ir.sia.cn/handle/173321/10762
Appears in Collections:光电信息技术研究室_学位论文

Files in This Item:
File Name/ File Size Content Type Version Access License
基于GPU的实时图像处理方法研究.pdf(2113KB)----限制开放 联系获取全文

Recommended Citation:
刘正斌.基于GPU的实时图像处理方法研究.[硕士学位论文].中国科学院沈阳自动化研究所.2013
Service
Recommend this item
Sava as my favorate item
Show this item's statistics
Export Endnote File
Google Scholar
Similar articles in Google Scholar
[刘正斌]'s Articles
CSDL cross search
Similar articles in CSDL Cross Search
[刘正斌]‘s Articles
Related Copyright Policies
Null
Social Bookmarking
Add to CiteULike Add to Connotea Add to Del.icio.us Add to Digg Add to Reddit
所有评论 (0)
暂无评论
 
评注功能仅针对注册用户开放,请您登录
您对该条目有什么异议,请填写以下表单,管理员会尽快联系您。
内 容:
Email:  *
单位:
验证码:   刷新
您在IR的使用过程中有什么好的想法或者建议可以反馈给我们。
标 题:
 *
内 容:
Email:  *
验证码:   刷新

Items in IR are protected by copyright, with all rights reserved, unless otherwise indicated.

 

 

Valid XHTML 1.0!
Copyright © 2007-2016  中国科学院沈阳自动化研究所 - Feedback
Powered by CSpace